Improving Disaster Responsiveness using a Mix

of Social Media and e-Government

Asanee Kawtrakul

1,2

, Intiraporn Mulasastra

, Hutchatai Chanlekha

, Sachit Rajbhandari

Kulapramote Prathumchai

, Masahiko Nagai

and Vasuthep Khunthong

Department of Computer Engineering, Kasetsart University, Bangkok, Thailand

National Electronic and Computer Technology Center, Bangkok, Thailand

Food and Agriculture Organization of the United Nations, Rome, Italy

Geoinformatics Center, Asian Institute of Technology, Khlong Luang, Thailand

Keywords: Data Integration, Disaster Management, e-Government, Ontology, Social Media.

Abstract: Data sharing is essential for government agencies during disaster management as it requires high

collaborative efforts among various organizations. Recently, social media have been increasingly used

during the disasters for disseminating and receiving information to and from the public. By using social

media for communications, the government can receive real-time data from the public and from

organizations. The challenge lies in how to combine social media with government data, which is gathered

from multiple sources, in multiple formats using multiple terminologies. This paper focuses on how to

manage, integrate, and verify data acquired from multiple sources. The proposed model was designed by

using frame-based data collection and ontology-based data integration, combined with the effective use of

dynamic data from social media, with the aim of improving the disaster assistance.

1 INTRODUCTION

The recent flood disaster in Thailand, starting in

September and continuing until December, 2011 was

the country's worst flood disaster in the last fifty

years. It became apparent that activities associated

with the disaster response needed more effective

management. There were too many government

agencies responsible for disaster management,

which made it difficult for collaboration and

coordination. As a result, flood water management

was poor, and flood relief goods and services were

not distributed equally to affected people. In

addition, there was conflicting and contradictory

information about the flooding. Many people,

including ordinary citizens and businesses, turned to

social media for accurate and up-to-date information

(Perry, 2011; Russell, 2011).

In order to achieve a higher level of efficiency

and accountability in the handling of disaster

response activities, accurate and up-to-date data are

crucially needed in the support of government

decision making. One of the key success factors in

managing response activities is access to real time

data, which could be provided by the local

population living in affected areas via social media.

The challenge lies in how to handle the mixed

social media and government data gathered from

multiple sources, in multiple formats, using multiple

terminologies. This paper focuses on how to manage

dynamic data and various sources of government

data. Using a flood response scenario as a case

study, a design of frame-based metadata for data

collection, and ontology-based data integration are

proposed.

2 CHALLENGES IN USING A

MIX OF SOCIAL MEDIA AND

E-GOVERNMENT

Providing assistance during disasters in a timely

manner needs e-Government data, as well as real-

time data that can be provided by local residents in

affected areas via social media. This section gives

the overview of the challenges in using mix data

from various sources.

423

Kawtrakul A., Mulasastra I., Chanlekha H., Rajbhandari S., Prathumchai K., Nagai M. and Khunthong V..

Improving Disaster Responsiveness using a Mix of Social Media and e-Government.

DOI: 10.5220/0004173404230426

In Proceedings of the International Conference on Knowledge Engineering and Ontology Development (KEOD-2012), pages 423-426

ISBN: 978-989-8565-30-3

 2012 SCITEPRESS (Science and Technology Publications, Lda.)

2.1 Disaster Response Management

The challenges of disaster response management are

the efficiency and responsiveness of assistance

provisions and supply distribution. These challenges

are:

 How to determine the location and number of

victims requiring assistance.

 How to extract and verify information of

assistance requests from social media.

 How to unify assistance efforts among public

and private organizations in providing

support.

 How to integrate data from back office

databases of government agencies with the

real time data from social media.

 How to estimate the severity of risk

management and determine the level of

required assistance.

2.2 Data Interoperability

During any crisis, data integration of related

organizations is essential. However, integrating the

data of various organizations is not easy. The

difficulties arise from the following problems.

2.2.1 Heterogeneous Data

The problems mentioned in the previous section are

those that need supporting data coming from various

government agencies. These data are heterogeneous

in nature. These heterogeneities can be classified as

syntactic, schematic and semantic (Bishr, 1998):

 Syntactic heterogeneity is caused by the use of

different data models, different file formats,

etc.

 Schematic heterogeneity is the result of

structural differences, caused by storing the

same data in different ways. For example, the

same information could be stored in a single

table or distributed in multiple tables, etc.

 Semantic heterogeneity is caused when the

same data has different meanings, or the same

data is interpreted differently under different

contexts. A semantic issue can arise when the

same term represents two different concepts,

or two terms represent the same concept.

Similarly, for spatial data, governmental

organizations usually run a variety of spatial

information systems. Different organizations usually

use different GIS software to handle spatial data in

different formats.

Since flood-related data is increasing in volume

and diversity, unifying metadata is insufficient to

enable valuable data to be accessed or understood.

One solution is to develop common metadata

standards and a metadata registry.

2.2.2 Data Quality

Although social media contains a lot of valuable

information, it is problematic with regard to quality

and needs to be carefully evaluated. Some of the

problems are: rumors, imprecise information,

contradictory information, and non standardized

information.

In social media, people write messages in non

formal languages without any standard. They may

describe flood depth in various ways or refer to a

location using the informal name that they are

accustomed to. Without standardization, it is

difficult to utilize such data effectively.

Because of the problem of having no standard,

we need both a knowledge base and a methodology

to normalize extracted information. Ontology can be

used to map locations reported with different names

and granularity to a normalized spatial system (e.g.

latitude-longitude, etc.). It also allows for more

flexible aggregation and reasoning in terms of

spatial information. For flood level, we need to

normalize a different measuring system into the

same system.

3 PROPOSED SOLUTIONS

During the disaster response phase, information

flows are very important in providing disaster

assistance (Day et al., 2009). However, there are

many issues related to acquiring, integrating and

verifying various data supporting operations. This

section describes the proposed solutions.

3.1 The Conceptual Framework

Figure 1 shows the conceptual framework, which

illustrates how data from different sources can be

extracted, integrated, harmonized and verified in

order to support disaster assistance management.

3.2 Social Media Information

Acquisition and Extraction

Despite the impurity or erroneous aspects of its

nature, social media still holds high potential to be a

useful data source for monitoring, prioritizing

KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

424

Figure 1: Analysis framework for the role of social media in disaster assistance management.

disaster severity, and managing resources for

disaster response and recovery. We can extract the

needed information from social media using NLP

techniques, as explained briefly below.

 Information Retrieval

This task could be performed by utilizing a social

media application interface (API), such as a Twitter

API, a Facebook graph API, as well as a Facebook

query language

 Information Filtering and Classification

Some of the retrieved messages may not be

related to a help request or flood report. We need a

means to filter irrelevant information and also

automatically classify messages into help requests

and reports. This can be done by using text

classification method (Sriram, et al., 2010).

 Information Extraction

Information extraction techniques can be used to

automatically extract predefined information and

structuralizing it in XML or database format.

3.3 Data Verification

Before using help demand data received from

various sources, e.g., social media and telephones,

we need to assess the level of data quality

(McGilvray, 2008) as follows:

 Data completeness

Data can be accessed from the completeness of

required data either in a single field or a combination

of required fields. If incomplete, then additional

sources of information are needed.

 Data accuracy

Data accuracy can be assessed by verifying with

reference sources. For example, the number of

people in a specific area can be verified with the

summary of the people registered in the system for

that area, or with mobile usage statistics from mobile

network operators.

 Data Reliability

Data source reliability is important for

eliminating malicious requests; it can be assessed by

given prioritization rules. For instance, if a source of

data is from an official, then the data sent can be

regarded as having a high level of reliability.

3.4 Harmonization and Integration

The syntactic issue raised due to multiple data

formats can be resolved by transforming them into a

single standard format, such as XML.

3.4.1 Data Harmonization

For semantic differences, data standards can be used

for harmonizing data. As recommended in our

previous study (Kawtrakul, et al., 2011), Thailand

has not established data standards at the national

level as yet. To prepare for and respond to disasters

that might occur in the future, the following

processes need to be prepared:

 Establishing universal core data standards.

ImprovingDisasterResponsivenessusingaMixofSocialMediaande-Government

425

 Forming a collaborative group which consists

of the representatives from all agencies that

are involved in providing help to disaster

victims. This group should create disaster

responsive common or domain specific data

standards.

3.4.2 Ontology based Data Integration

An ontology is an explicit specification of a

conceptualization. Ontology can be used to support

semantic data sharing and data integration among

these organizations (Blomqvist and Öhgren, 2008).

Hence, the concept of ontology is recommended to

support data sharing and integration.

Semantic data integration based on ontology uses

a conceptual representation of the data and their

relationships to solve heterogeneities problems. A

Simple Knowledge Organization System (Miles and

Bechhofer, 2009), provides a common data model

for sharing and linking knowledge organization

systems. In SKOS, ontology can be represented in

such a way that concepts can be shared among

organizations which are involved in data sharing and

integration.

3.4.3 Gazetteer based Spatial Data

Integration

Water level estimates cannot be monitored by

satellite remote sensing. Flood depths are collected

by social network and integrated with spatial data

such as DEM (Digital Elevation Model), flood

extent maps, etc. Flood depth is point information

that is provided in meters with latitude and

longitude. This point data from social networks

usually identifies the location by geographical

names, such as city name, village name, landmark,

etc. Gazetteer is utilized to convert geographical

names to latitude and longitude (Nagai and Ono,

2008). Geographical point data is analyzed and

interpolated to grid data which will then be overlaid

and integrated with spatial data so as to support

flood level estimates.

4 CONCLUSIONS AND FUTURE

WORK

When a disaster occurs, the critical performance of

disaster relief operations is timely responsiveness.

Since social media has been proven to be a major

means of disseminating and receiving information

during a crisis, social media mixed with e-

Government creates a new level of data integration

for responding to help requests. Data required during

a crisis are from various sources in different formats.

Hence, it is essential for government to prepare the

technology for integrating and harmonizing

information systems for disaster management. To

handle disaster situations, the government should

also consider the following issues:

 Establishment of a national standards body

that is accountable for supporting data

interoperability and data standardization.

 Creation of a disaster collaborative group that

consists of all government agencies, the

military, and private organizations that need to

collaborate in exchanging data during a crisis.

REFERENCES

Bishr, Y., 1998. Overcoming the semantic and other

barriers to GIS interoperability, Int. J. of Geographical

Inform. Sci., 12(4), pp.229–314.

Blomqvist, E., Öhgren, A., 2008. Constructing an

enterprise ontology for an automotive supplier,

Journal Engineering Application of Artificial

Intelligence, 21(3), pp. 386-397.

Day, J. M., Junglas, I., Silva, L., 2009. Information Flow

Impediments in Disaster Relief Supply Chains, J. of

Assoc. for Inform. Syst., 10(8), pp. 637-660.

Kawtrakul, A., Mulasastra, I., Khampachua, T.,

Ruengittinun, S., 2011. Moving Fast Forward to

National Data Standardization, In: 11th European

Conference on eGovernment (ECEG11). Ljubljana,

Slovenia 16-17 June 2011.

Miles, A., Bechhofer, S., 2009. Simple Knowledge

Organization System. [Online] W3C. Available at: <

http://www.w3.org/TR/skos-reference/> [Accessed 10

January 2012].

Nagai, M., Ono, M., 2008. Ontology Development for

Interoperability of Observation Data, In: XXI

International Society for Photogrammetry and Remote

Sensing (ISPRS). Beijing, China 3-11 Jul 2008.

Perry, B., 2011. Social Media Innovation Flourishes

During Thailand Floods, [Online] TECHINASIA.

Available at: <http://www.techinasia.com/thailand-

flood-social-media-innovation/> [Accessed 1 April

2012].

Russell, J., 2011. As floods batter Thailand, its citizens

turn to the Internet for help, [Online] TNW:Asia.

Available at: <http://thenextweb.com/asia/ 2011/10/

14/as-floods-batter-thailand-its-citizens-turn-to-the-

internet-for-help/> [Accessed 1 April 2012].

Sriram, B. Fuhry, D., Demir, E., Ferhatosmanoglu, H.,

Demirbas, M., 2010. Short text classification in twitter

to improve information filtering, In: 33rd Int. ACM

SIGIR Conference on Research and Development in

Information Retrieval. Geneva, Switzerland 19-23 July

2010.

KEOD2012-InternationalConferenceonKnowledgeEngineeringandOntologyDevelopment

426